broker-dealers and perhaps a dozen clearing agencies, which is what I interpret Mr. Peake is suggesting.

Mr. PAINTER. But isn't there a need for a uniform system? Or let me put the question a little differently. Is there a risk that if you have a number of different automated systems being developed, possibly one by the Midwest Stock Exchange, one by the Pacific Coast Stock Exchange, one by the New York Stock Exchange, one by the National Clearing Corporation, that if you have these various systems being developed and they are all automated, is there a risk that they may not be able to interface somewhere along the line?

In other words, don't you need a comprehensive national system eventually somewhere down the road or at least some kind of standards, or you may be getting into problems of incompatibility of very sophisticated automated systems?

Mr. WEINBERG. Yes, I think you need the standards which you described and which Chairman Moss has alluded to in questions to Mr. Sporkin. We need the entity to design the system and develop the standards.

I think that entity is what is important and is needed but I think we can separate that from the role of actually operating the system. I think it is critical that we make that distinction.

Mr. PAINTER. You recommend also that this organization be financed preferably by means of a minimal charge levied on all security transactions. This just raised in my mind the question, Why should the cost of restructuring the securities industry be the burden of the investing public? Wouldn't it be more appropriate for the industry to finance solutions to their own problems?

Mr. WEINBERG. Let me state that I think the kind of changes we are talking about here would result in a dramatic reduction in the cost to the smaller investor. Suggestions like eliminating the stock certificate or improving the handling of it, I think there is general agreement that will result in reducing the commission charged by as much as 30, 40, or 50 percent.

The offsetting costs that we talk about to support this kind of an agency to guide the development would probably be talking about adding literally 5 or 10 cents per trade, so the relative tradeoff between the savings and the cost are minimal. The extra costs will be small.

The question of how we assess the charge, because there has to be a charge, is an open question. We think by putting it on each transaction, in a sense like a transfer tax, we don't then have to resolve the problem of where the trade is going to take place, or what entity is going to be responsible for collection.

We know that every change of ownership has to go through a transfer agent. That seems like an appropriate way to assess the tax and to collect it. We don't want to inhibit people from trading on the exchange, for example. So we don't want to put it on as an exchange tax.

But the important thing is the order of magnitude, the amount of fees that we would have to collect. The amount of tax would be negligible in relation to the saving to the investor.

Mr. PAINTER. I would like to move on a little to the question of the transition to the certificateless society, if we are going to have one, and I have a question here for Mr. Noyes.

On pages G-2 and G-3 of your statement, Mr. Noyes, you discuss two ways in which to proceed as far as implementation of an automated system is concerned, and the first would require that any changeover would be made by a series of small steps.

The second would establish a completely automated system in its entirety, and then begin by switching stock issues into that system a few at a time until the old system was replaced.

I wondered which of these two alternatives seems to be the preferable one in your opinion.

Mr. NOYES. From a technical point of view, perhaps the latter, the completely new system developed switched in, much as the issues handled by CCS were switched in one issue at a time.

Perhaps, though, considering the problem of easing the transition and getting agreement from all parts of the industry, maybe we need an evolution. I am going to beg the question. I really don't know what the answer is to the best way of getting the job done. Technically, I would like to see the system built from the ground up and then move into it.

Mr. PAINTER. I notice, Mr. Weinberg, on page 7 of your statement, you express the opinion that the development of a certificateless system should not involve any sudden wholesale conversion to a new system. Do you have any comments with regard to the two possible alternatives that Mr. Noyes has suggested?

Mr. WEINBERG. Primarily we are trying to put to rest a fear that at some date somebody is going to be asked to send in all of their stock certificates and we are going to lock them in a vault.

We don't think that will be necessary. A stratification by amount of activity will result in change where it is most needed relatively quickly, and those shares which seldom change hands would be brought in on a phase basis when they do trade.

So, stratification of the implementation by type of trader gives you a kind of phase-in transition period.

Mr. PAINTER. I suppose it would be entirely appropriate that this new Federal body that is being suggested consider, among its various problems, this one of the best way to make a transition to a certificateless society?

Mr. WEINBERG. Yes.

Mr. PAINTER. Does any other panelist have comments he would like to make on this problem of making a transition?

Mr. MALLER. This would be something that would be obviously a part of the study to be done in the specific design and implementation of a system. However, I visualize two potentials.

One would be the conversion of zone by zone as outlined in my concent which is a matter of record.

The second is the depositories, which are a very valid and very helpful conversion tool, would be converted one at a time with the development of various depositories throughout the country such as the Kassenvereine in Germany today. These two exist as alternate routes, and again the specific implementation is one of the jobs designers have while they are doing the study for design of the system. Mr. PAINTER. Of course, it is obvious that you first have to make a decision what the system will be before you decide how you are going to make the transition.

Shall I move on to the problem of the cost of developing a certificateless society? I note that on pages 2 and 11 of Mr. Maller's statement, he estimates the total cost of developing and implementing the Fastclear system as $300 million with annual operating cost of $70 million.

In rather sharp contrast, the North American Rockwell figures submitted on page C-3 of its statement run much lower, $2 million to $3 million a year for the first year of system design and a total cost of only $40 million to $50 million.

Also, Mr. Weinberg in his statement, page 14, mentions a figure of "at least $15 million."

What factors lie behind these rather significant discrepancies and what makes the Fastclear estimates so much higher than the estimates for the system proposed by North American Rockwell? Mr. Maller, would you like to comment?

Mr. MALLER. Yes, I have made certain assumptions so that I would have the widest flexibility possible in a system. One fundamental assumption was that every person in the United States could conceivably make one transaction each day. This would mean that we would have 400 million transactions to be entered daily into the system inasmuch as it takes two entries for each transaction.

The other fundamental assumption is that every person in the United States has 100 holdings and that each of these holdings can be expressed in 100 characters of identification.

Mr. PAINTER. This would be a very fortunate state of affairs if all this were so.

Mr. MALLER. Well, again my concept is that we must have a storage capacity and working file in the main data base which would be capable of flexibility to almost infinite numbers.

Based on these, I then made some calculations as to the storage rates of various machine capacities, the kinds of compression that could be made. I then included a factor of 10, increased all of our calculations by a factor of 10, to give us the opportunity to have additional flexibility and expansion if need be.

I went into the question of the machine hierarchy and programs and that we have to be able to edit, verify, and back up all of the data and from that I moved into the programing costs.

Programing costs I assumed would be comparable to the development of a comprehensive operating system. I used the development of IBM OS as an example. This took 1.000 man-years of work, 250 men working 4 years. This would cost about $30 million, considering an average cost here of $30,000 a man. This includes implementation.

I then looked at the question of equipment and the cost of equipment to handle the kinds of bytes of data I came up with, and this brought me to total cost of $300 million. I hope it is excessive.

Mr. PAINTER. Mr. Noyes, do you have any comments to make? I know you have already given us an estimate of about $40 to $50 million to develop the entire system. Do you consider that to be an accurate one in view of the differing figures which have been submitted. I suppose they are all admittedly educated guesses.

Mr. NOYES. Yes, exactly. I think there are a couple of basic differences between the estimates that Mr. Maller has made and those we did. One is that the systems are quite different. As I understand the system that he is suggesting, it is a clearance system whereas, after

the trade is made, the parties to the trade submit data about the trade. These are cleared, and then the records are kept at the beneficial owner level.

Also, he has suggested a rather higher capacity for the system than we had estimated, and, thirdly, our estimate was based on additional cost over and above the capability that exists today, such as communication links, computing capacity at the firms and exchanges, and so on. So this was an additional cost. I suspect that his was a total systems

cost.

Mr. PAINTER. On page C-4 of your statement you refer to the possibility that the estimates might be "low by a factor of two or three" if each step should require manual monitoring and control. Will you explain what you mean by that statement?

Mr. NOYES. Yes, I think again probably all of us envisage a slightly different system or maybe substantially different system. I address many of my responses to not only the settlement process, but the trading process, and there are different ways that trading can be done.

One example is, or one way the trading can be done is, for many trades to be done completely automatically, such as the odd lot and single round lot systems being proposed now by New York and American Exchanges and perhaps that idea could be carried on into multiple round lot transactions.

Alternatively, to build a trading system that has a high degree of manual intervention, people working in it, or tying in a NASDAQ terminal, complicates the problem considerably. So to the extent those are in the system and have to be designed and built in, the cost will increase.

Mr. PAINTER. When you say a factor of about two or three, you mean it may be twice as expensive or possibly three times as expensive? Mr. NOYES. Yes.

Mr. PAINTER. On page a-1 of your statement you observe the computing capacity required by an automated stock transaction system is not particularly large and that the implementation of such a system would not require as much information or expense as systems like American Airlines' SABRE system, the Air Force's SAGE system, or the computer and ground communications networks required to support the Apollo missions.

I wondered whether any of the other panelists have a reaction to this statement. Is it accurate? Secondly, if American Airlines has done this, why can't this be done by the securities industry and done fairly quickly?

Mr. SPORKIN. Mr. Painter, I get the feeling here that we are talking about designing a completely new system, and I feel that there have been tremendous accomplishments and there are existing systems that all they need to be is adapted, I mentioned the NASDAQ system, and have talked about the over-the-counter market, but I don't want to get involved in choosing what the ultimate marketplace will look like. The NASDAQ system has already put a communication system in every firm in the United States. I think everybody will agree that is one of the most important things that any system can be, a communication device.

Secondly, you now have the NCC which is going to come on line within the next 60 to 90 days. And, by the way, NCC, as I recall, only cost about $2 to $3 million.

Mr. PAINTER. $2 to $3 million?

Mr. SPORKIN. I think $3 million is high. I think it was more around $2 million which was financed by NASD itself.

Now, I am not a computer man or systems man, but I am trying to give you the best I can from being in this business for a long period of time. If you had a NASDAQ system which could be adapted to execute transactions, and, as I see execution, it would execute with a locked-in trade, which everybody agreed last week is the important thing so you know you have a trade, you don't have to go back and try to find out whether you did or did not have a trade.

Mr. PAINTER. You are implying that NASDAQ could be so adapted?

Mr. SPORKIN. Yes, the NASDAQ system is now set up so that you can now quote securities and communicate between firms. I don't know how much it would take to turn that into an execution type of system, at least for over-the-counter end of it, that would mean that broker A could communicate with broker B. He could get the quote, and, second, he could say, "We buy."

They would both then put an input into this terminal which would be linked to the main computer in Connecticut, where both sides of the transaction would be stored.

Now, unless that transaction locked, it would be immediately sent back to the broker and reported: "You don't have a trade," or, "You do have a trade," so immediately you would know whether you have a good trade.

You tie that into the NCC development that they are going to have and you, in addition to NCC, you now have these depositories that have developed through CCS and what have you.

So, if you had NCC, with a depository system which would, in effect, immobilize the certificate, plus a transaction effecting system which could be easily adapted, I believe, from the system you now have, you may well be going 75 to 80 percent down the line with developing

a system.

Now, I would pose this question to the technical people here and ask them if what I am saying makes any sense or not as to being able to adapt what you already have at cost which would be, perhaps, onetenth of the cost that we are talking about, and do it quickly?

Mr. PAINTER. Does any panelist have any comments to make to Mr. Sporkin's remarks?

Mr. WEINBERG. I agree with the concept Mr. Sporkin has described, and in addition to NASDAQ, there are other examples.

The First National Bank of Boston has developed a transfer system which can operate either with or without the existence of physical certificates.

They have now instituted, in conjunction with the Boston Stock Clearing Corporation certificateless transfers for a local Boston company, a publishing house. We find these kinds of systems have been de veloped. A good many of the necessary elements for a certification system already exist. I don't think we have to start from scratch in developing a brand new certificateless clearing system. I think it should build on what already exists.

Mr. PAINTER. Is the Commission encouraging this development which you describe, Mr. Sporkin?